Data
impact_data <- read_spss("data/cleandata/final_imPACT_long_desire_stress_jessica_nick.sav")
impact_data_long <- read_spss("data/cleandata/15_05_14_imPACT_use_for_big_T.sav")
# replace NaN with NA
impact_data[is.na(impact_data)] <- NA
impact_data_long[is.na(impact_data_long)] <- NA
Join with uncentered BMI
impact_data <- impact_data %>%
left_join(
impact_data_long %>%
distinct(ID, BMI_B), by = "ID"
)
Change and/or create new variables
impact_data <- impact_data %>%
mutate(gender_chr = case_when(Gender == 0 ~ "Man", Gender == 1 ~ "Woman"),
gender_fct = factor(gender_chr),
relationship_status_chr = case_when(relstatcoded == 1 ~ "Single", relstatcoded == 2 ~ "Casual", relstatcoded == 3 ~ "Committed"),
relationship_status_fct = factor(relationship_status_chr, levels = c("Single", "Casual", "Committed")),
nicotine_chr = case_when(nicotine == 0 ~ "No", relstatcoded == 1 ~ "Yes"),
nicotine_fct = factor(nicotine_chr),
sex_med_chr = case_when(sexmeds == 0 ~ "No", sexmeds == 1 ~ "Yes"),
sex_med_fct = factor(sex_med_chr),
testosterone_med_chr = case_when(TMeds == 0 ~ "No", TMeds == 1 ~ "Yes"),
testosterone_med_fct = factor(testosterone_med_chr),
cortisol_med_chr = case_when(CMeds == 0 ~ "No", CMeds == 1 ~ "Yes"),
cortisol_med_fct = factor(cortisol_med_chr),
hormonal_contraceptive_chr = case_when(HC == 0 ~ "No", HC == 1 ~ "Yes"),
hormonal_contraceptive_fct = factor(hormonal_contraceptive_chr),
illness_chr = case_when(illness == 0 ~ "No", illness == 1 ~ "Cold or minor includes unspecified", illness == 2 ~ "More intense or chronic"),
illness_fct = factor(illness_chr, levels = c("No", "Cold or minor includes unspecified", "More intense or chronic")),
relationship_status_fct = factor(relationship_status_chr))
Set contrasts
contrasts(impact_data$relationship_status_fct) <- contr.poly(n = 3)
contrasts(impact_data$illness_fct) <- contr.poly(n = 3)
contrasts(impact_data$hormonal_contraceptive_fct) <- c(-1, 1)
Subset data by exclusion criteria and gender
# men and women
impact_data_nodrugs <- impact_data %>%
filter(nicotine_chr == "No" & sex_med_chr == "No" & testosterone_med_chr == "No" & cortisol_med_chr == "No") %>%
mutate(time_since_waking_dbl = parse_number(as.character(timesincewake)),
time_since_waking_c = time_since_waking_dbl - mean(time_since_waking_dbl, na.rm = TRUE),
bmi_c = BMI_B - mean(BMI_B, na.rm = TRUE),
testosterone_screen_c = Testosteronescreen - mean(Testosteronescreen, na.rm = TRUE),
cortisol_screen_c = Cortscreen - mean(Cortscreen, na.rm = TRUE),
pss_c = PSS - mean(PSS, na.rm = TRUE))
# men
impact_data_nodrugs_men <- impact_data %>%
filter(gender_chr == "Man" & nicotine_chr == "No" & sex_med_chr == "No" & testosterone_med_chr == "No" & cortisol_med_chr == "No") %>%
mutate(time_since_waking_dbl = parse_number(as.character(timesincewake)),
time_since_waking_c = time_since_waking_dbl - mean(time_since_waking_dbl, na.rm = TRUE),
bmi_c = BMI_B - mean(BMI_B, na.rm = TRUE),
testosterone_screen_c = Testosteronescreen - mean(Testosteronescreen, na.rm = TRUE),
cortisol_screen_c = Cortscreen - mean(Cortscreen, na.rm = TRUE),
pss_c = PSS - mean(PSS, na.rm = TRUE))
# women
impact_data_nodrugs_women <- impact_data %>%
filter(gender_chr == "Woman" & nicotine_chr == "No" & sex_med_chr == "No" & testosterone_med_chr == "No" & cortisol_med_chr == "No") %>%
mutate(time_since_waking_dbl = parse_number(as.character(timesincewake)),
time_since_waking_c = time_since_waking_dbl - mean(time_since_waking_dbl, na.rm = TRUE),
bmi_c = BMI_B - mean(BMI_B, na.rm = TRUE),
testosterone_screen_c = Testosteronescreen - mean(Testosteronescreen, na.rm = TRUE),
cortisol_screen_c = Cortscreen - mean(Cortscreen, na.rm = TRUE),
pss_c = PSS - mean(PSS, na.rm = TRUE))
Exploratory Plots
Sum Dyadic Sexual Desire (6-62 points) by gender (boxplots)
impact_data_nodrugs %>%
ggplot(mapping = aes(x = ordered(Time_AdjZero), y = SDI_dyad, fill = gender_fct)) +
geom_boxplot() +
scale_x_discrete(breaks = seq(0, 11, 1)) +
scale_y_continuous(breaks = seq(0, 70, 10)) +
scale_fill_manual(values = c("#0072B2", "#CC79A7")) +
labs(x = "Session (0-11 session numbers)", y = NULL, title = "Sum Dyadic Sexual Desire (6-62 points) by gender", fill = NULL) +
theme(legend.position = "top")
## Warning: Removed 25 rows containing non-finite values (stat_boxplot).
Sum Dyadic Sexual Desire (6-62 points) by gender (cubic splines)
impact_data_nodrugs %>%
ggplot(mapping = aes(x = Time_AdjZero, y = SDI_dyad, fill = gender_fct, color = gender_fct)) +
geom_smooth(method = "gam", formula = y ~ s(x, bs = "cs", k = 5), se = FALSE) +
geom_point(alpha = 0.25, size = 3) +
scale_x_continuous(breaks = seq(0, 11, 1)) +
scale_y_continuous(breaks = seq(0, 70, 10)) +
scale_fill_manual(values = c("#0072B2", "#CC79A7")) +
scale_color_manual(values = c("#0072B2", "#CC79A7")) +
labs(x = "Session (0-11 session numbers)", y = NULL, title = "Sum Dyadic Sexual Desire (6-62 points) by gender", subtitle = "Cubic Spline Smooth, k = 5", fill = NULL, color = NULL) +
theme(legend.position = "top")
## Warning: Removed 25 rows containing non-finite values (stat_smooth).
## Warning: Removed 25 rows containing missing values (geom_point).
Sum Solitary Sexual Desire (2-23 points) by gender (boxplots)
impact_data_nodrugs %>%
ggplot(mapping = aes(x = ordered(Time_AdjZero), y = SDI_sol, fill = gender_fct)) +
geom_boxplot() +
scale_x_discrete(breaks = seq(0, 11, 1)) +
scale_y_continuous(breaks = seq(0, 25, 5)) +
scale_fill_manual(values = c("#0072B2", "#CC79A7")) +
labs(x = "Session (0-11 session numbers)", y = NULL, title = "Sum Solitary Sexual Desire (2-23 points) by gender", fill = NULL) +
theme(legend.position = "top")
## Warning: Removed 23 rows containing non-finite values (stat_boxplot).
Sum Solitary Sexual Desire (2-23 points) by gender (cubic splines)
impact_data_nodrugs %>%
ggplot(mapping = aes(x = Time_AdjZero, y = SDI_sol, fill = gender_fct, color = gender_fct)) +
geom_smooth(method = "gam", formula = y ~ s(x, bs = "cs", k = 5), se = FALSE) +
geom_point(alpha = 0.25, size = 3) +
scale_x_continuous(breaks = seq(0, 11, 1)) +
scale_y_continuous(breaks = seq(0, 25, 5)) +
scale_fill_manual(values = c("#0072B2", "#CC79A7")) +
scale_color_manual(values = c("#0072B2", "#CC79A7")) +
labs(x = "Session (0-11 session numbers)", y = NULL, title = "Sum Solitary Sexual Desire (2-23 points) by gender", subtitle = "Cubic Spline Smooth, k = 5", fill = NULL, color = NULL) +
theme(legend.position = "top")
## Warning: Removed 23 rows containing non-finite values (stat_smooth).
## Warning: Removed 23 rows containing missing values (geom_point).
Relationship among testosterone, cortisol, and dyadic sexual desire in men
impact_data_nodrugs_men %>%
mutate(cortisol_screen_c_q = cut(cortisol_screen_c, breaks = quantile(cortisol_screen_c, probs = seq(0, 1, 1 / 5), na.rm = TRUE), include.lowest = TRUE)) %>%
filter(!is.na(cortisol_screen_c_q)) %>%
ggplot(mapping = aes(x = testosterone_screen_c, y = SDI_dyad)) +
geom_smooth(method = "gam", formula = y ~ s(x, bs = "cs"), se = FALSE, color = "#0072B2") +
geom_point(alpha = 0.25, size = 3, color = "#0072B2") +
facet_wrap(facets = ~ cortisol_screen_c_q) +
scale_y_continuous(breaks = seq(0, 70, 10)) +
labs(x = "Testosterone (mean-centered)", y = NULL, title = "Sum Dyadic Sexual Desire (6-62 points) by mean-centered cortisol quintile in men") +
theme(legend.position = "top")
## Warning: Removed 17 rows containing non-finite values (stat_smooth).
## Warning: Removed 17 rows containing missing values (geom_point).
Relationship among testosterone, cortisol, and solitary sexual desire in men
impact_data_nodrugs_men %>%
mutate(cortisol_screen_c_q = cut(cortisol_screen_c, breaks = quantile(cortisol_screen_c, probs = seq(0, 1, 1 / 5), na.rm = TRUE), include.lowest = TRUE)) %>%
filter(!is.na(cortisol_screen_c_q)) %>%
ggplot(mapping = aes(x = testosterone_screen_c, y = SDI_sol)) +
geom_smooth(method = "gam", formula = y ~ s(x, bs = "cs"), se = FALSE, color = "#0072B2") +
geom_point(alpha = 0.25, size = 3, color = "#0072B2") +
facet_wrap(facets = ~ cortisol_screen_c_q) +
scale_y_continuous(breaks = seq(0, 25, 5)) +
labs(x = "Testosterone (mean-centered)", y = NULL, title = "Sum Solitary Sexual Desire (2-23 points) by mean-centered cortisol quintile in men") +
theme(legend.position = "top")
## Warning: Removed 20 rows containing non-finite values (stat_smooth).
## Warning: Removed 20 rows containing missing values (geom_point).
Relationship among testosterone, cortisol, and dyadic sexual desire in women
impact_data_nodrugs_women %>%
mutate(cortisol_screen_c_q = cut(cortisol_screen_c, breaks = quantile(cortisol_screen_c, probs = seq(0, 1, 1 / 5), na.rm = TRUE), include.lowest = TRUE)) %>%
filter(!is.na(cortisol_screen_c_q)) %>%
ggplot(mapping = aes(x = testosterone_screen_c, y = SDI_dyad)) +
geom_smooth(method = "gam", formula = y ~ s(x, bs = "cs"), se = FALSE, color = "#CC79A7") +
geom_point(alpha = 0.25, size = 3, color = "#CC79A7") +
facet_wrap(facets = ~ cortisol_screen_c_q) +
scale_y_continuous(breaks = seq(0, 70, 10)) +
labs(x = "Testosterone (mean-centered)", y = NULL, title = "Sum Dyadic Sexual Desire (6-62 points) by mean-centered cortisol quintile in women") +
theme(legend.position = "top")
## Warning: Removed 22 rows containing non-finite values (stat_smooth).
## Warning: Removed 22 rows containing missing values (geom_point).
Relationship among testosterone, cortisol, and solitary sexual desire in women
impact_data_nodrugs_women %>%
mutate(cortisol_screen_c_q = cut(cortisol_screen_c, breaks = quantile(cortisol_screen_c, probs = seq(0, 1, 1 / 5), na.rm = TRUE), include.lowest = TRUE)) %>%
filter(!is.na(cortisol_screen_c_q)) %>%
ggplot(mapping = aes(x = testosterone_screen_c, y = SDI_sol)) +
geom_smooth(method = "gam", formula = y ~ s(x, bs = "cs"), se = FALSE, color = "#CC79A7") +
geom_point(alpha = 0.25, size = 3, color = "#CC79A7") +
facet_wrap(facets = ~ cortisol_screen_c_q) +
scale_y_continuous(breaks = seq(0, 25, 5)) +
labs(x = "Testosterone", y = NULL, title = "Sum Solitary Sexual Desire (2-23 points) by mean-centered cortisol quintile in women") +
theme(legend.position = "top")
## Warning: Removed 17 rows containing non-finite values (stat_smooth).
## Warning: Removed 17 rows containing missing values (geom_point).
Relationship among testosterone, perceived stress, and dyadic sexual desire in men
impact_data_nodrugs_men %>%
mutate(pss_c_q = cut(pss_c, breaks = quantile(pss_c, probs = seq(0, 1, 1 / 5), na.rm = TRUE), include.lowest = TRUE)) %>%
filter(!is.na(pss_c_q)) %>%
ggplot(mapping = aes(x = testosterone_screen_c, y = SDI_dyad)) +
geom_smooth(method = "gam", formula = y ~ s(x, bs = "cs"), se = FALSE, color = "#0072B2") +
geom_point(alpha = 0.25, size = 3, color = "#0072B2") +
facet_wrap(facets = ~ pss_c_q) +
scale_y_continuous(breaks = seq(0, 70, 10)) +
labs(x = "Testosterone (mean-centered)", y = NULL, title = "Sum Dyadic Sexual Desire (6-62 points) by mean-centered perceived stress quintile in men") +
theme(legend.position = "top")
## Warning: Removed 27 rows containing non-finite values (stat_smooth).
## Warning: Removed 27 rows containing missing values (geom_point).
Relationship among testosterone, perceived stress, and solitary sexual desire in men
impact_data_nodrugs_men %>%
mutate(pss_c_q = cut(pss_c, breaks = quantile(pss_c, probs = seq(0, 1, 1 / 5), na.rm = TRUE), include.lowest = TRUE)) %>%
filter(!is.na(pss_c_q)) %>%
ggplot(mapping = aes(x = testosterone_screen_c, y = SDI_sol)) +
geom_smooth(method = "gam", formula = y ~ s(x, bs = "cs"), se = FALSE, color = "#0072B2") +
geom_point(alpha = 0.25, size = 3, color = "#0072B2") +
facet_wrap(facets = ~ pss_c_q) +
scale_y_continuous(breaks = seq(0, 25, 5)) +
labs(x = "Testosterone (mean-centered)", y = NULL, title = "Sum Solitary Sexual Desire (2-23 points) by mean-centered perceived stress quintile in men") +
theme(legend.position = "top")
## Warning: Removed 31 rows containing non-finite values (stat_smooth).
## Warning: Removed 31 rows containing missing values (geom_point).
Relationship among testosterone, perceived stress, and dyadic sexual desire in women
impact_data_nodrugs_women %>%
mutate(pss_c_q = cut(pss_c, breaks = quantile(pss_c, probs = seq(0, 1, 1 / 5), na.rm = TRUE), include.lowest = TRUE)) %>%
filter(!is.na(pss_c_q)) %>%
ggplot(mapping = aes(x = testosterone_screen_c, y = SDI_dyad)) +
geom_smooth(method = "gam", formula = y ~ s(x, bs = "cs"), se = FALSE, color = "#CC79A7") +
geom_point(alpha = 0.25, size = 3, color = "#CC79A7") +
facet_wrap(facets = ~ pss_c_q) +
scale_y_continuous(breaks = seq(0, 70, 10)) +
labs(x = "Testosterone (mean-centered)", y = NULL, title = "Sum Dyadic Sexual Desire (6-62 points) by mean-centered perceived stress quintile in women") +
theme(legend.position = "top")
## Warning: Removed 36 rows containing non-finite values (stat_smooth).
## Warning: Removed 36 rows containing missing values (geom_point).
Relationship among testosterone, perceived stress, and solitary sexual desire in women
impact_data_nodrugs_women %>%
mutate(pss_c_q = cut(pss_c, breaks = quantile(pss_c, probs = seq(0, 1, 1 / 5), na.rm = TRUE), include.lowest = TRUE)) %>%
filter(!is.na(pss_c_q)) %>%
ggplot(mapping = aes(x = testosterone_screen_c, y = SDI_sol)) +
geom_smooth(method = "gam", formula = y ~ s(x, bs = "cs"), se = FALSE, color = "#CC79A7") +
geom_point(alpha = 0.25, size = 3, color = "#CC79A7") +
facet_wrap(facets = ~ pss_c_q) +
scale_y_continuous(breaks = seq(0, 25, 5)) +
labs(x = "Testosterone", y = NULL, title = "Sum Solitary Sexual Desire (2-23 points) by mean-centered perceived stress quintile in women") +
theme(legend.position = "top")
## Warning: Removed 31 rows containing non-finite values (stat_smooth).
## Warning: Removed 31 rows containing missing values (geom_point).
Models
Fit model 1
lmer_fit1 <- lmer(SDI_dyad ~ bmi_c + time_since_waking_c + relationship_status_fct + illness_fct + Time_AdjZero * (testosterone_screen_c * cortisol_screen_c + testosterone_screen_c * pss_c) + (1 + Time_AdjZero | ID), data = impact_data_nodrugs_men, REML = TRUE)
## Warning: Some predictor variables are on very different scales: consider
## rescaling
## Warning: Some predictor variables are on very different scales: consider
## rescaling
Diagnostic Plots
# residuals vs. fitted
plot(lmer_fit1, type = c("p", "smooth"))
# random effects
dotplot.ranef.mer(ranef(lmer_fit1, condVar = TRUE))
## $ID
Summary
summary(lmer_fit1)
## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: SDI_dyad ~ bmi_c + time_since_waking_c + relationship_status_fct +
## illness_fct + Time_AdjZero * (testosterone_screen_c * cortisol_screen_c +
## testosterone_screen_c * pss_c) + (1 + Time_AdjZero | ID)
## Data: impact_data_nodrugs_men
##
## REML criterion at convergence: 1536.6
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -2.29207 -0.52799 0.01013 0.47190 2.90016
##
## Random effects:
## Groups Name Variance Std.Dev. Corr
## ID (Intercept) 64.103 8.006
## Time_AdjZero 1.003 1.002 -0.31
## Residual 25.458 5.046
## Number of obs: 222, groups: ID, 62
##
## Fixed effects:
## Estimate Std. Error
## (Intercept) 3.342e+01 3.168e+00
## bmi_c 1.445e-01 2.752e-01
## time_since_waking_c -4.703e-02 1.178e-01
## relationship_status_fct.L -4.670e+00 1.187e+00
## relationship_status_fct.Q -5.268e+00 1.310e+00
## illness_fct.L -1.079e+01 6.105e+00
## illness_fct.Q -5.470e+00 3.594e+00
## Time_AdjZero -1.598e-01 2.448e-01
## testosterone_screen_c -5.509e-02 3.417e-02
## cortisol_screen_c 1.113e+00 8.467e-01
## pss_c 1.306e-01 1.109e-01
## testosterone_screen_c:cortisol_screen_c 1.639e-03 1.948e-02
## testosterone_screen_c:pss_c 4.235e-03 3.645e-03
## Time_AdjZero:testosterone_screen_c 1.969e-02 7.626e-03
## Time_AdjZero:cortisol_screen_c -2.260e-01 1.986e-01
## Time_AdjZero:pss_c 2.296e-02 3.068e-02
## Time_AdjZero:testosterone_screen_c:cortisol_screen_c -5.342e-04 3.776e-03
## Time_AdjZero:testosterone_screen_c:pss_c -3.997e-04 7.930e-04
## df t value
## (Intercept) 9.437e+01 10.549
## bmi_c 5.947e+01 0.525
## time_since_waking_c 1.464e+02 -0.399
## relationship_status_fct.L 1.802e+02 -3.934
## relationship_status_fct.Q 1.940e+02 -4.022
## illness_fct.L 7.226e+01 -1.767
## illness_fct.Q 8.020e+01 -1.522
## Time_AdjZero 2.053e+01 -0.653
## testosterone_screen_c 1.815e+02 -1.612
## cortisol_screen_c 1.403e+02 1.315
## pss_c 1.571e+02 1.178
## testosterone_screen_c:cortisol_screen_c 1.748e+02 0.084
## testosterone_screen_c:pss_c 1.754e+02 1.162
## Time_AdjZero:testosterone_screen_c 1.283e+02 2.583
## Time_AdjZero:cortisol_screen_c 1.125e+02 -1.138
## Time_AdjZero:pss_c 1.198e+02 0.748
## Time_AdjZero:testosterone_screen_c:cortisol_screen_c 1.225e+02 -0.141
## Time_AdjZero:testosterone_screen_c:pss_c 1.444e+02 -0.504
## Pr(>|t|)
## (Intercept) < 2e-16 ***
## bmi_c 0.601369
## time_since_waking_c 0.690274
## relationship_status_fct.L 0.000119 ***
## relationship_status_fct.Q 8.27e-05 ***
## illness_fct.L 0.081491 .
## illness_fct.Q 0.131902
## Time_AdjZero 0.521315
## testosterone_screen_c 0.108676
## cortisol_screen_c 0.190726
## pss_c 0.240658
## testosterone_screen_c:cortisol_screen_c 0.933045
## testosterone_screen_c:pss_c 0.246809
## Time_AdjZero:testosterone_screen_c 0.010927 *
## Time_AdjZero:cortisol_screen_c 0.257654
## Time_AdjZero:pss_c 0.455768
## Time_AdjZero:testosterone_screen_c:cortisol_screen_c 0.887734
## Time_AdjZero:testosterone_screen_c:pss_c 0.615001
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation matrix not shown by default, as p = 18 > 12.
## Use print(x, correlation=TRUE) or
## vcov(x) if you need it
## fit warnings:
## Some predictor variables are on very different scales: consider rescaling
Variance Partition Coefficients
lmer_vpc(lmer_fit1)
## # A tibble: 4 x 7
## grp var1 var2 vcov sdcor vpc vpc_pct
## <chr> <chr> <chr> <dbl> <dbl> <dbl> <chr>
## 1 ID (Intercept) <NA> 64.1 8.01 0.728 72.76%
## 2 ID Time_AdjZero <NA> 1.00 1.00 0.0114 1.14%
## 3 ID (Intercept) Time_AdjZero -2.46 -0.307 -0.0280 -2.8%
## 4 Residual <NA> <NA> 25.5 5.05 0.289 28.9%
Fit model 2
lmer_fit2 <- lmer(SDI_sol ~ bmi_c + time_since_waking_c + relationship_status_fct + illness_fct + Time_AdjZero * (testosterone_screen_c * cortisol_screen_c + testosterone_screen_c * pss_c) + (1 + Time_AdjZero | ID), data = impact_data_nodrugs_men, REML = TRUE)
## Warning: Some predictor variables are on very different scales: consider
## rescaling
## Warning: Some predictor variables are on very different scales: consider
## rescaling
Diagnostic Plots
# residuals vs. fitted
plot(lmer_fit2, type = c("p", "smooth"))
# random effects
dotplot.ranef.mer(ranef(lmer_fit2, condVar = TRUE))
## $ID
Summary
summary(lmer_fit2)
## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: SDI_sol ~ bmi_c + time_since_waking_c + relationship_status_fct +
## illness_fct + Time_AdjZero * (testosterone_screen_c * cortisol_screen_c +
## testosterone_screen_c * pss_c) + (1 + Time_AdjZero | ID)
## Data: impact_data_nodrugs_men
##
## REML criterion at convergence: 1209.5
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -3.02331 -0.46265 -0.06901 0.38740 3.13362
##
## Random effects:
## Groups Name Variance Std.Dev. Corr
## ID (Intercept) 18.398 4.2893
## Time_AdjZero 0.128 0.3578 -0.41
## Residual 5.323 2.3071
## Number of obs: 219, groups: ID, 62
##
## Fixed effects:
## Estimate Std. Error
## (Intercept) 7.495e+00 1.378e+00
## bmi_c 2.114e-01 1.411e-01
## time_since_waking_c 5.665e-02 5.361e-02
## relationship_status_fct.L -4.090e-01 5.508e-01
## relationship_status_fct.Q -7.621e-01 6.280e-01
## illness_fct.L -2.830e+00 2.557e+00
## illness_fct.Q -2.565e+00 1.516e+00
## Time_AdjZero 1.441e-01 9.663e-02
## testosterone_screen_c 1.043e-02 1.586e-02
## cortisol_screen_c 7.885e-01 3.873e-01
## pss_c 5.504e-02 5.149e-02
## testosterone_screen_c:cortisol_screen_c 2.787e-03 8.922e-03
## testosterone_screen_c:pss_c -2.536e-03 1.742e-03
## Time_AdjZero:testosterone_screen_c -2.429e-05 3.347e-03
## Time_AdjZero:cortisol_screen_c -1.362e-01 8.792e-02
## Time_AdjZero:pss_c -1.650e-02 1.342e-02
## Time_AdjZero:testosterone_screen_c:cortisol_screen_c -1.008e-03 1.657e-03
## Time_AdjZero:testosterone_screen_c:pss_c 2.663e-05 3.527e-04
## df t value
## (Intercept) 8.781e+01 5.440
## bmi_c 5.896e+01 1.498
## time_since_waking_c 1.472e+02 1.057
## relationship_status_fct.L 1.786e+02 -0.743
## relationship_status_fct.Q 1.893e+02 -1.213
## illness_fct.L 5.942e+01 -1.107
## illness_fct.Q 6.851e+01 -1.692
## Time_AdjZero 2.703e+01 1.491
## testosterone_screen_c 1.730e+02 0.657
## cortisol_screen_c 1.401e+02 2.036
## pss_c 1.601e+02 1.069
## testosterone_screen_c:cortisol_screen_c 1.642e+02 0.312
## testosterone_screen_c:pss_c 1.678e+02 -1.455
## Time_AdjZero:testosterone_screen_c 1.100e+02 -0.007
## Time_AdjZero:cortisol_screen_c 9.947e+01 -1.549
## Time_AdjZero:pss_c 1.066e+02 -1.230
## Time_AdjZero:testosterone_screen_c:cortisol_screen_c 1.030e+02 -0.608
## Time_AdjZero:testosterone_screen_c:pss_c 1.312e+02 0.076
## Pr(>|t|)
## (Intercept) 4.76e-07 ***
## bmi_c 0.1395
## time_since_waking_c 0.2924
## relationship_status_fct.L 0.4587
## relationship_status_fct.Q 0.2265
## illness_fct.L 0.2728
## illness_fct.Q 0.0953 .
## Time_AdjZero 0.1475
## testosterone_screen_c 0.5118
## cortisol_screen_c 0.0436 *
## pss_c 0.2868
## testosterone_screen_c:cortisol_screen_c 0.7552
## testosterone_screen_c:pss_c 0.1474
## Time_AdjZero:testosterone_screen_c 0.9942
## Time_AdjZero:cortisol_screen_c 0.1244
## Time_AdjZero:pss_c 0.2216
## Time_AdjZero:testosterone_screen_c:cortisol_screen_c 0.5443
## Time_AdjZero:testosterone_screen_c:pss_c 0.9399
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation matrix not shown by default, as p = 18 > 12.
## Use print(x, correlation=TRUE) or
## vcov(x) if you need it
## fit warnings:
## Some predictor variables are on very different scales: consider rescaling
Variance Partition Coefficients
lmer_vpc(lmer_fit2)
## # A tibble: 4 x 7
## grp var1 var2 vcov sdcor vpc vpc_pct
## <chr> <chr> <chr> <dbl> <dbl> <dbl> <chr>
## 1 ID (Intercept) <NA> 18.4 4.29 0.792 79.24%
## 2 ID Time_AdjZero <NA> 0.128 0.358 0.00551 0.55%
## 3 ID (Intercept) Time_AdjZero -0.630 -0.411 -0.0271 -2.71%
## 4 Residual <NA> <NA> 5.32 2.31 0.229 22.92%
Fit model 3
lmer_fit3 <- lmer(SDI_dyad ~ bmi_c + time_since_waking_c + relationship_status_fct + illness_fct + hormonal_contraceptive_fct + Time_AdjZero * (testosterone_screen_c * cortisol_screen_c + testosterone_screen_c * pss_c) + (1 + Time_AdjZero | ID), data = impact_data_nodrugs_women, REML = TRUE)
Diagnostic Plots
# residuals vs. fitted
plot(lmer_fit3, type = c("p", "smooth"))
# random effects
dotplot.ranef.mer(ranef(lmer_fit3, condVar = TRUE))
## $ID
Summary
summary(lmer_fit3)
## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: SDI_dyad ~ bmi_c + time_since_waking_c + relationship_status_fct +
## illness_fct + hormonal_contraceptive_fct + Time_AdjZero *
## (testosterone_screen_c * cortisol_screen_c + testosterone_screen_c *
## pss_c) + (1 + Time_AdjZero | ID)
## Data: impact_data_nodrugs_women
##
## REML criterion at convergence: 1786.9
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -2.60643 -0.43524 0.07023 0.50614 2.61812
##
## Random effects:
## Groups Name Variance Std.Dev. Corr
## ID (Intercept) 127.5322 11.2930
## Time_AdjZero 0.9864 0.9932 -0.47
## Residual 18.3164 4.2798
## Number of obs: 267, groups: ID, 62
##
## Fixed effects:
## Estimate Std. Error
## (Intercept) 30.488306 1.806817
## bmi_c 0.654744 0.299189
## time_since_waking_c 0.064576 0.091562
## relationship_status_fct.L -1.349718 0.814299
## relationship_status_fct.Q -4.624644 1.149390
## illness_fct.L -3.853232 1.835839
## illness_fct.Q -2.826055 1.290266
## hormonal_contraceptive_fct1 1.139215 0.961691
## Time_AdjZero 0.102374 0.209092
## testosterone_screen_c -0.338364 0.149392
## cortisol_screen_c -0.498657 0.621110
## pss_c -0.062762 0.118676
## testosterone_screen_c:cortisol_screen_c 0.083385 0.093879
## testosterone_screen_c:pss_c -0.024002 0.020258
## Time_AdjZero:testosterone_screen_c 0.003839 0.030564
## Time_AdjZero:cortisol_screen_c 0.259325 0.142796
## Time_AdjZero:pss_c -0.024481 0.024435
## Time_AdjZero:testosterone_screen_c:cortisol_screen_c -0.018275 0.020608
## Time_AdjZero:testosterone_screen_c:pss_c 0.004724 0.003851
## df t value
## (Intercept) 102.443497 16.874
## bmi_c 62.121612 2.188
## time_since_waking_c 172.604791 0.705
## relationship_status_fct.L 184.799372 -1.658
## relationship_status_fct.Q 244.096781 -4.024
## illness_fct.L 166.995058 -2.099
## illness_fct.Q 176.686919 -2.190
## hormonal_contraceptive_fct1 222.509003 1.185
## Time_AdjZero 41.617307 0.490
## testosterone_screen_c 215.440675 -2.265
## cortisol_screen_c 207.821220 -0.803
## pss_c 211.625613 -0.529
## testosterone_screen_c:cortisol_screen_c 207.982612 0.888
## testosterone_screen_c:pss_c 200.254853 -1.185
## Time_AdjZero:testosterone_screen_c 168.614509 0.126
## Time_AdjZero:cortisol_screen_c 189.541394 1.816
## Time_AdjZero:pss_c 146.635226 -1.002
## Time_AdjZero:testosterone_screen_c:cortisol_screen_c 186.792922 -0.887
## Time_AdjZero:testosterone_screen_c:pss_c 155.218831 1.227
## Pr(>|t|)
## (Intercept) < 2e-16 ***
## bmi_c 0.0324 *
## time_since_waking_c 0.4816
## relationship_status_fct.L 0.0991 .
## relationship_status_fct.Q 7.65e-05 ***
## illness_fct.L 0.0373 *
## illness_fct.Q 0.0298 *
## hormonal_contraceptive_fct1 0.2374
## Time_AdjZero 0.6270
## testosterone_screen_c 0.0245 *
## cortisol_screen_c 0.4230
## pss_c 0.5975
## testosterone_screen_c:cortisol_screen_c 0.3754
## testosterone_screen_c:pss_c 0.2375
## Time_AdjZero:testosterone_screen_c 0.9002
## Time_AdjZero:cortisol_screen_c 0.0709 .
## Time_AdjZero:pss_c 0.3180
## Time_AdjZero:testosterone_screen_c:cortisol_screen_c 0.3763
## Time_AdjZero:testosterone_screen_c:pss_c 0.2218
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation matrix not shown by default, as p = 19 > 12.
## Use print(x, correlation=TRUE) or
## vcov(x) if you need it
Variance Partition Coefficients
lmer_vpc(lmer_fit3)
## # A tibble: 4 x 7
## grp var1 var2 vcov sdcor vpc vpc_pct
## <chr> <chr> <chr> <dbl> <dbl> <dbl> <chr>
## 1 ID (Intercept) <NA> 128. 11.3 0.901 90.07%
## 2 ID Time_AdjZero <NA> 0.986 0.993 0.00697 0.7%
## 3 ID (Intercept) Time_AdjZero -5.24 -0.467 -0.0370 -3.7%
## 4 Residual <NA> <NA> 18.3 4.28 0.129 12.94%
Fit model 4
lmer_fit4 <- lmer(SDI_sol ~ bmi_c + time_since_waking_c + relationship_status_fct + illness_fct + hormonal_contraceptive_fct + Time_AdjZero * (testosterone_screen_c * cortisol_screen_c + testosterone_screen_c * pss_c) + (1 + Time_AdjZero | ID), data = impact_data_nodrugs_women, REML = TRUE)
Diagnostic Plots
# residuals vs. fitted
plot(lmer_fit4, type = c("p", "smooth"))
# random effects
dotplot.ranef.mer(ranef(lmer_fit4, condVar = TRUE))
## $ID
Summary
summary(lmer_fit4)
## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: SDI_sol ~ bmi_c + time_since_waking_c + relationship_status_fct +
## illness_fct + hormonal_contraceptive_fct + Time_AdjZero *
## (testosterone_screen_c * cortisol_screen_c + testosterone_screen_c *
## pss_c) + (1 + Time_AdjZero | ID)
## Data: impact_data_nodrugs_women
##
## REML criterion at convergence: 1243.5
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -4.5128 -0.2872 -0.0430 0.3133 3.0994
##
## Random effects:
## Groups Name Variance Std.Dev. Corr
## ID (Intercept) 21.74688 4.6634
## Time_AdjZero 0.01928 0.1389 0.05
## Residual 1.78059 1.3344
## Number of obs: 272, groups: ID, 62
##
## Fixed effects:
## Estimate Std. Error
## (Intercept) 5.271079 0.682179
## bmi_c 0.616864 0.131623
## time_since_waking_c 0.025806 0.027857
## relationship_status_fct.L 0.438282 0.252444
## relationship_status_fct.Q -0.241520 0.342597
## illness_fct.L 0.033198 0.562770
## illness_fct.Q 0.148842 0.393304
## hormonal_contraceptive_fct1 -0.455631 0.318846
## Time_AdjZero 0.013051 0.046156
## testosterone_screen_c -0.077775 0.042873
## cortisol_screen_c 0.112399 0.176193
## pss_c 0.016828 0.033914
## testosterone_screen_c:cortisol_screen_c 0.048831 0.027259
## testosterone_screen_c:pss_c -0.014175 0.005756
## Time_AdjZero:testosterone_screen_c 0.021953 0.008258
## Time_AdjZero:cortisol_screen_c -0.057791 0.038724
## Time_AdjZero:pss_c -0.006305 0.006422
## Time_AdjZero:testosterone_screen_c:cortisol_screen_c -0.004872 0.005640
## Time_AdjZero:testosterone_screen_c:pss_c 0.003125 0.001024
## df t value
## (Intercept) 88.498836 7.727
## bmi_c 60.100235 4.687
## time_since_waking_c 180.743414 0.926
## relationship_status_fct.L 185.586781 1.736
## relationship_status_fct.Q 151.603194 -0.705
## illness_fct.L 173.719644 0.059
## illness_fct.Q 188.885413 0.378
## hormonal_contraceptive_fct1 239.213891 -1.429
## Time_AdjZero 18.438617 0.283
## testosterone_screen_c 177.764141 -1.814
## cortisol_screen_c 164.053622 0.638
## pss_c 171.475846 0.496
## testosterone_screen_c:cortisol_screen_c 182.800334 1.791
## testosterone_screen_c:pss_c 175.413774 -2.463
## Time_AdjZero:testosterone_screen_c 88.288384 2.658
## Time_AdjZero:cortisol_screen_c 75.440968 -1.492
## Time_AdjZero:pss_c 57.475074 -0.982
## Time_AdjZero:testosterone_screen_c:cortisol_screen_c 92.153371 -0.864
## Time_AdjZero:testosterone_screen_c:pss_c 78.731656 3.051
## Pr(>|t|)
## (Intercept) 1.62e-11 ***
## bmi_c 1.64e-05 ***
## time_since_waking_c 0.35548
## relationship_status_fct.L 0.08420 .
## relationship_status_fct.Q 0.48191
## illness_fct.L 0.95303
## illness_fct.Q 0.70553
## hormonal_contraceptive_fct1 0.15431
## Time_AdjZero 0.78051
## testosterone_screen_c 0.07135 .
## cortisol_screen_c 0.52441
## pss_c 0.62039
## testosterone_screen_c:cortisol_screen_c 0.07489 .
## testosterone_screen_c:pss_c 0.01476 *
## Time_AdjZero:testosterone_screen_c 0.00932 **
## Time_AdjZero:cortisol_screen_c 0.13977
## Time_AdjZero:pss_c 0.33030
## Time_AdjZero:testosterone_screen_c:cortisol_screen_c 0.38988
## Time_AdjZero:testosterone_screen_c:pss_c 0.00311 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation matrix not shown by default, as p = 19 > 12.
## Use print(x, correlation=TRUE) or
## vcov(x) if you need it
Variance Partition Coefficients
lmer_vpc(lmer_fit4)
## # A tibble: 4 x 7
## grp var1 var2 vcov sdcor vpc vpc_pct
## <chr> <chr> <chr> <dbl> <dbl> <dbl> <chr>
## 1 ID (Intercept) <NA> 21.7 4.66 0.922 92.23%
## 2 ID Time_AdjZero <NA> 0.0193 0.139 0.000818 0.08%
## 3 ID (Intercept) Time_AdjZero 0.0326 0.0503 0.00138 0.14%
## 4 Residual <NA> <NA> 1.78 1.33 0.0755 7.55%